Exploratory Factor Analysis (EFA)

Một phần của tài liệu Perceptions of University Digital Libraries as information source by international postgraduate student (Trang 188 - 197)

The next subsections describe the EFA performed in the study. EFA was utilised to assess the construct validity of the following constructs: System Features and Internal Differences measured as per the study’s conceptual model (Table 4.18). As mentioned earlier, only these two constructs are scrutinised as it was believed that they are the most relevant in the context of the present study.

Table 4.17 Constructs Included in the Conceptual Framework

Construct Sub-Construct

Individual Differences

Computer Self-efficacy (SE) Computer Experience (CS) Domain Knowledge (DK) Motivation (M0)

System Features

Accessibility (AC) Visibility (VI) Relevance (RE)

4.7.2.1 EFA for System Features (Accessibility, Visibility, Relevance)

The Kaiser-Meyer-Olkin (KMO) test and the Bartlett’s test of sphericity are typically utilised to ascertain the factorability of the output matrix of a scale (Coakes, 2005; Pallant, 2005). In general, the feasibility of the factor analysis is indicated by high values of the KMO test (>0.50; de Vaus, 2002; Field, 2005; Netemeyer, Bearden, & Sharma, 2003) and high significance value of the Bartlett’s test. The KMO Measure of Sampling Adequacy, with a value of 0.866, indicates that the sample size was sufficiently large to perform factor analysis for the System Features construct.

Moreover, the Bartlett’s test of sphericity was significant with p=0.000, indicating adequate correlations between the variables (Table 4.19).

Table 4.18 KMO and Bartlett's Test for System Features

Kaiser-Meyer-Olkin Measure of Sampling Adequacy. 0.866

Bartlett's Test of Sphericity

Approx. Chi-Square 4031.604

df 78

Sig. 0.000

189 The outcomes of the factor analysis for the System Features construct are provided in Table 4.20.

Factors with eigenvalues of >1 and a factor loading of at least 0.5 were considered acceptable. It can be seen that the facets related to the Accessibility of a system was the most important factor that could explain 49.697% of the variance in system features, followed by Visibility and Relevance. Moreover, it could be seen that all the items in each construct had factor values greater than the cut-off level.

Table 4.19 Factors of System Features Variable

Code Factors Factor

loadings

% of Variance

Cumulative

%

Accessibility 49.697 49.697

AC1 I find it easy to navigate 0.846

AC2 I am able to use it whenever I need it 0.828

AC3 I find it easy to get access to 0.859

AC4 It is easily accessible 0.773

AC5 I can locate the resources I need 0.848

Visibility 16.385 66.082

VI1 People at my university know that it exists 0.869 VI2 People know where to look to find it 0.855

VI3 I find that it is always available 0.740

Relevance 8.371 74.453

RE1 It has resources that relate to my area of interest 0.739 RE2 It has enough resources for my study 0.845 RE3 It provides current information in my area of

interest 0.540

RE4 It is a very efficient study tool 0.510

RE5 It is limited in its coverage of my area of interest 0.886

The EFA for the UDL dataset is described next.

UDL Dataset

The KMO Measure of Sampling Adequacy, with a value of 0.753, indicates that the sample size was sufficiently large to perform factor analysis for the System Features construct in the UDL

190 dataset. Moreover, the Bartlett’s test of sphericity was significant with p=0.000, indicating adequate correlations between the variables (Table 4.21).

Table 4.20 KMO and Bartlett’s test for System Features – UDL Dataset Kaiser-Meyer-Olkin Measure of Sampling Adequacy. 0.753 Bartlett's Test of Sphericity Approx. Chi-Square 1411.543

df 78

Sig. 0.000

The outcomes of the factor analysis for the System Features construct are provided in Table 4.22.

Factors with eigenvalues of >1 and a factor loading of at least 0.5 were considered acceptable. In contrast to the combined dataset, it can be seen that the facets related to the Relevance of a system was the most important factor that could explain 38.003% of the variance in system features, followed by Accessibility and Visibility. Further, it can be seen that all the items in each construct had factor values greater than the cut-off level.

Table 4.21 Factors of System Features – UDL Dataset Variable

Code Factors Factor Loadings % of

Variance

Cumulative

%

Relevance 38.003 38.003

RE1 It has resources that relate to my area

of interest 0.778

RE2 It has enough resources for my study 0.827

RE3 It provides current information in my

area of interest 0.675

RE4 It is a very efficient study tool 0.511

RE5 It is limited in its coverage of my area

of interest 0.875

Accessibility 15.526 53.530

AC1 I find it easy to navigate 0.817

191 Variable

Code Factors Factor Loadings % of

Variance

Cumulative

%

AC2 I am able to use it whenever I need it 0.732

AC3 I find it easy to get access to 0.864

AC4 It is easily accessible 0.711

AC5 I can locate the resources I need 0.814

Visibility 13.414 66.944

VI1 People at my university know that it

exists 0.871

VI2 People know where to look to find it 0.900

VI3 I find that it is always available 0.709

Google Scholar Dataset

The KMO Measure of Sampling Adequacy, with a value of 0.800, indicates that the sample size was sufficiently large to perform factor analysis for the System Features construct in the Google Scholar dataset. Moreover, the Bartlett’s test of sphericity was significant with p=0.000, indicating adequate correlations between the variables (Table 4.23).

Table 4.22 KMO and Bartlett’s Test for System Features - Google Scholar Dataset Kaiser-Meyer-Olkin Measure of Sampling Adequacy. 0.800 Bartlett's Test of Sphericity Approx. Chi-Square 1383.825

df 55

Sig. 0.000

The outcomes of the factor analysis for the System Features construct are provided in Table 4.24.

Factors with eigenvalues of >1 and a factor loading of at least 0.5 were considered acceptable.

Similar to the UDL dataset, it can be seen that the facets related to the Relevance of a system was the most important factor that could explain 45.619% of the variance in system features, followed by Accessibility and Relevance. Moreover, it can be seen that all the items in each construct had factor values greater than the cut-off level.

192 Table 4.23 Factors of System Features - Google Scholar Dataset

Variable Code

Factors Factor Loadings % of

Variance

Cumulative

%

Relevance 45.619 45.619

RE1 It has resources that relate to my

area of interest 0.741

RE2 It has enough resources for my study 0.817 RE5 It is limited in its coverage of my

area of interest 0.858

Accessibility 17.741 63.360

AC1 I find it easy to navigate 0.891

AC2 I am able to use it whenever I need

it 0.795

AC3 I find it easy to get access to 0.866

AC4 It is easily accessible 0.727

AC5 I can locate the resources I need 0.833

Visibility 10.881 74.241

VI1 People at my university know that it

exists 0.877

VI2 People know where to look to find it 0.823 VI3 I find that it is always available 0.829

4.7.2.2 EFA for Internal Differences (Domain Knowledge, Computer Experience, Computer Self-efficacy, Motivation)

The KMO Measure of Sampling Adequacy, with a value of 0.791, indicates that the sample size was sufficiently large to perform factor analysis for the Internal Differences construct. Moreover, the Bartlett’s test of sphericity was significant with p=0.000, indicating adequate correlations between the variables (Table 4.25).

Table 4.24 KMO and Bartlett’s Test for Internal Differences

Kaiser-Meyer-Olkin Measure of Sampling Adequacy. 0.791

Bartlett's Test of Sphericity

Approx. Chi-Square 4433.648

df 171

Sig. 0.000

193 The outcomes of the factor analysis for the Internal Differences construct are provided in Table 4.26. Factors with eigenvalues of >1 and a factor loading of at least 0.5 were considered acceptable.

It can be seen that the facets related to the Domain Knowledge of an individual was the most important factor that could explain 31.259% of the variance in internal differences, this was followed by Motivation, Computer Self-efficacy, and Computer Experience. Moreover, it could be seen that all the items in each construct had factor values greater than the cut-off level.

Table 4.25 Factors of Internal Differences

Variable Code Factors Factor loadings % of

Variance

Cumulative

%

Domain Knowledge 31.259 31.259

DK1 I am familiar with the subject

domain that I search for 0.880 DK2 I am knowledgeable in the topic to

search for 0.894

DK3 I have previous experience

searching in this subject domain 0.848 DK4

I have the domain knowledge that it necessary to search for what I want to find

0.840

Motivation 14.069 45.328

MO1 Helps me achieve in my studies 0.861 MO2 I use it because people around me do 0.726

MO3 I have been trained to use it 0.762

MO4 I am confident in using it 0.457

MO5 I don’t always feel in control of the

outcome 0.798

MO6 Makes me feel really involved in my

studies 0.456

Computer Self-Efficacy 11.198 56.526

SE1 I feel confident in my ability to use

it 0.794

SE2 I can use it even if there is no one

around me to show me 0.693

SE3 I don’t need a lot of time to complete

my task using it 0.767

SE4 I often find it difficult to use it for

my studies 0.659

194

Variable Code Factors Factor loadings % of

Variance

Cumulative

% SE5 Helps even when the task is

challenging 0.767

Computer Experience 8.334 64.860

CS1 I am confident in using computers 0.800 CS2 I think I am efficient in the use of a

computer to complete my task 0.900 CS3 I can use a computer even if there is

no one around to show me 0.872 CS4 I am happier if there is someone

around to ask for help 0.431 UDL Dataset

The KMO Measure of Sampling Adequacy, with a value of 0.675, indicates that the sample size was sufficiently large to perform factor analysis for the Internal Differences construct in the UDL dataset. Moreover, the Bartlett’s test of sphericity was significant with p=0.000, indicating adequate correlations between the variables (Table 4.27).

Table 4.26 KMO and Bartlett’s test for Individual Differences – UDL Dataset Kaiser-Meyer-Olkin Measure of Sampling Adequacy. 0.675 Bartlett's Test of Sphericity 1639.049 1364.857

136 153

0.000 0.000

The outcomes of the factor analysis for the Internal Differences construct are provided in Table 4.28. Factors with eigenvalues of >1 and a factor loading of at least 0.5 were considered acceptable.

It could be seen that the facets related to the Domain Knowledge of an individual was the most important factor that could explain 22.222% of the variance in internal differences, this was followed by Computer Experience, Motivation, and Computer Self-efficacy. In contrast to the combined dataset, the factor loadings of items CS4 and MO6 did not meet the cut-off and could be excluded from further analysis.

195 Table 4.27 Factors of Individual Differences – UDL Dataset

Variable Code Factors

Factor Loadin gs

% of Varian ce

Cumul ative %

Domain Knowledge 22.222 22.222

DK1 I am familiar with the subject domain that I search

for 0.800

DK2 I am knowledgeable in the topic to search for 0.801 DK3 I have previous experience searching in this subject

domain 0.760

DK4 I have the domain knowledge that it necessary to

search for what I want to find 0.720

Computer Experience 16.897 39.120

CS1 I am confident in using computers 0.826

CS2 I think I am efficient in the use of a computer to

complete my task 0.956

CS3 I can use a computer even if there is no one around

to show me 0.926

Motivation 13.501 52.621

MO1 Helps me achieve in my studies 0.871

MO2 I use it because people around me do 0.649

MO3 I have been trained to use it 0.699

MO4 I am confident in using it 0.541

MO5 I don’t always feel in control of the outcome 0.813

Computer Self-Efficacy 10.178 62.799

SE1 I feel confident in my ability to use it 0.812

SE2 I can use it even if there is no one around me to

show me 0.711

SE3 I don’t need a lot of time to complete my task using

it 0.754

SE4 I often find it difficult to use it for my studies 0.657

SE5 Helps even when the task is challenging 0.719

Google Scholar Dataset

The KMO Measure of Sampling Adequacy, with a value of 0.669, indicates that the sample size was sufficiently large to perform factor analysis for the Internal Differences construct in the Google Scholar dataset. Moreover, the Bartlett’s test of sphericity was significant with p=0.000, indicating adequate correlations between the variables (Table 4.29).

196 Table 4.28 KMO and Bartlett’s test for Individual Differences – Google Scholar Dataset

Kaiser-Meyer-Olkin Measure of Sampling Adequacy. 0.669 Bartlett's Test of Sphericity Approx. Chi-Square 1364.857

df 153

Sig. 0.000

The outcomes of the factor analysis for the Internal Differences construct are provided in Table 4.30. Factors with eigenvalues of >1 and a factor loading of at least 0.5 were considered acceptable.

It could be seen that the facets related to the Domain Knowledge of an individual was the most important factor that could explain 21.297% of the variance in internal differences, this was followed by Computer Experience, Computer Self-efficacy, and Motivation. In contrast to the combined dataset, the factor loadings of item MO4 did not meet the cut-off and could be excluded from further analysis.

Table 4.29 Factors of Individual Differences – Google Scholar Dataset

Variable Code Factors Factor

Loadings

% of Variance

Cumul ative

%

Domain Knowledge 21.297 21.297

DK1 I am familiar with the subject domain that I search for

0.798

DK2 I am knowledgeable in the topic to search for 0.794

DK3 I have previous experience searching in this subject domain

0.774

DK4 I have the domain knowledge that it necessary

to search for what I want to find

0.715

Computer Experience 14.627 35.925

CS1 I am confident in using computers 0.689

CS2 I think I am efficient in the use of a computer to complete my task

0.839

CS3 I can use a computer even if there is no one around to show me

0.797

197

Variable Code Factors Factor

Loadings

% of Variance

Cumul ative

% CS4 I am happier if there is someone around to ask

for help

0.612

Computer Self-Efficacy 12.349 48.274

SE1 I feel confident in my ability to use it 0.787

SE2 I can use it even if there is no one around me to show me

0.681

SE3 I don’t need a lot of time to complete my task

using it

0.782

SE4 I often find it difficult to use it for my studies 0.618

SE5 Helps even when the task is challenging 0.741

Motivation 10.344 58.617

MO1 Helps me achieve in my studies 0.844

MO2 I use it because people around me do 0.752

MO3 I have been trained to use it 0.776

MO5 I don’t always feel in control of the outcome 0.800

MO6 Makes me feel really involved in my studies 0.526

The next section describes the CFA performed in the study in further detail.

Một phần của tài liệu Perceptions of University Digital Libraries as information source by international postgraduate student (Trang 188 - 197)

Tải bản đầy đủ (PDF)

(366 trang)